In fact, arterial endothelial cells are subjected to both fluid shear stress and cyclic hoop stretch in vivo. Therefore, a more complete investigation of mechanical.
Annals of Biomedical Engineering, Vol. 22, pp. 4 1 6 ~ 2 2 , 1994
0090-6964/94 $10.50 + .00 Copyright 9 1994 Biomedical Engineering Society
Printed in the USA. All rights reserved.
A Device for Subjecting Vascular Endothelial Cells to Both Fluid Shear Stress and Circumferential Cyclic Stretch JAMES E . M O O R E , J R . , * ERNST BLIRKI,t ANDREAS S U C I U , * SHUMIN Z H A O , t M I C H E L B U R N I E R , t H A N S R . B R U N N E R , t a n d JEAN-JACQUES MEISTER* *Biomedical Engineering Laboratory, Swiss Federal Institute of Technology and tHypertension Division, University Hospital, Lausanne, Switzerland
Abstract--The proposal of the role of mechanical forces as a localizing factor of atherosclerosis has led many researchers to investigate their effects on vascular endothelial cells. Most previous efforts have concentrated on either the fluid shear stress, which results from the flow of blood, or the circumferential "hoop" stretch, which results from the expansion of the artery during the cardiac cycle. In fact, arterial endothelial cells are subjected to both fluid shear stress and cyclic hoop stretch in vivo. Therefore, a more complete investigation of mechanical phenomena on endothelial cell behavior should include both kinds of mechanical stimuli. This study was undertaken to design an experimental apparatus that could subject cultured vascular endothelial cells to simultaneous physiologic levels of both shear stress and cyclic hoop stretch. The experimental apparatus consists of four cylindrical elastic tubes so that the following conditions may be studied: (a) static conditions; (b) shear stress only; (c) hoop stretch only; and (d) shear stress and hoop stretch. In order to establish the functional capabilities of the apparatus, bovine pulmonary artery endothelial cells were cultured in the tubes, and their morphology and f-actin structure were observed with confocal microscopy. The cells remained healthy and attached to the walls throughout the 24 hr experiment. Preliminary results indicated that the alignment of endothelial cells subjected to shear stress was significantly enhanced by the addition of hoop strain.
tery during the cardiac cycle. Much research has been devoted to the investigation of the effects of these factors on the cells that make up the artery wall. The effects of wall shear stress on laboratory cultured vascular endothelial cells have been extensively investigated by several groups. It has been noted that shear stress can significantly alter the morphology of the cells as well as the production of various substances (2). The elongation and alignment of the cells in the direction of flow is one often quoted effect of shear stress (6). Other research efforts have focused on the effects of cyclic stretching on endothelial cells. Since these cells are attached to the inner walls of arteries, they are subjected to a cyclic hoop stretch as the artery diameter increases during systole and decreases during diastole. The strain is normally less than 10% in large and medium-sized arteries. The effects on the morphology and function of endothelial cells subjected to cyclic stretching have also been well documented by different groups, and were recently summarized by Banes (1). As with shear stress, the cells are seen to elongate and align, but in this case perpendicular to the direction of stretching (5). In a cylindrical geometry, this means that the cells should align parallel to the axis of the tube. In fact, arterial endothelial cells are simultaneously subjected to both fluid shear stress and cyclic hoop stretch in vivo. Therefore, a more complete investigation of mechanical phenomena on endothelial cell behavior should include both shear stress and hoop stretch. This study was undertaken to design an experimental apparatus that could subject cultured vascular endothelial cells to both shear stress and cyclic hoop stretch. The design of the device from a mechanical and biological point of view is presented here, along with some preliminary results to demonstrate its performance.
Keywords--Endothelium, Shear stress, Cyclic stretch. INTRODUCTION Mechanical factors have been proposed to be involved in the formation of cardiovascular diseases such as atherosclerosis. The two types of mechanical factors that have received the most attention are the fluid shear stress that results from the flow of blood and the circumferential " h o o p " stretch that results from the expansion of the arAcknowledgment--Supported in part by the Swiss National Scientific Research Fund, grant no. 32-32535.91, and by the CHUV-EPFL-UNIL Biomedical Engineering Common Collaboration Program. The authors gratefully acknowledge the assistance of Mr. Joel L. Berry in the initial design of the apparatus. Address correspondenceto Dr. Moore, BiomedicalEngineering Laboratory, EPFL, Champs----courbes1, CH 1024 Ecublens, Switzerland.
METHODS
The design of an experimental apparatus that could subject cells to shear stress and hoop stretch must take into
(Received 13Dec93, Revised 22Feb94, Accepted 18Mar94)
416
Shearing and Stretching Endothelium account several important mechanical as well as biological factors. Mechanically speaking, the apparatus must provide physiologic levels of shear stress, while at the same time allowing the cells to be stretched. From a biological point of view, the cells must be in a sterile nourishing environment. The overall design of this apparatus is centered around four tubes, which constitute the study section (Fig. 1). The basic idea is to culture the cells on the inner walls of compliant tubes, then apply a pulsatile pressure gradient that imposes a pulsatile shear stress and a cyclic expansion on the tubes. Four tubes are included so that the following conditions may be studied: (a) static conditions (no shear stress and no hoop stretch); (b) shear stress only; (c) hoop stretch only; and (d) shear stress and hoop stretch. In this manner, the results from the most complex mechanical environment (d) may be compared with simpler cases (b and c), then with static conditions (a). For the tubes where no shear stress was desired, the flow rate was adjusted to be just enough to nourish the cells (20 ml/min, shear stress < 0.5 dyne/cm2). For the tubes where no hoop stretch was desired, a rigid plastic casing was mounted around the outsides of the compliant tubes to prevent their expansion. The compliant tubes were custom manufactured using Dow Coming Sylgard 184. A very thin layer ( - 0 . 2 mm) of the material was applied in liquid state to a highly polished 6 mm diameter cylinder that was constantly, slowly rotated in an oven at 150~ The application of liquid Sylgard to the mandrel was carefully controlled to ensure that the thickness of the tubes did not vary around the circumference or along the length. The tubes were
417
allowed to solidify inside the oven, then they were carefully removed from the mandrel. They were then mounted onto specially designed fittings that kept them at a constant length using two aluminum rods (Fig. 2). An initial axial stretch of 10% was used to ensure that the tubes remained straight under pressure. This stretch was kept constant throughout the culturing process and flow experiments. The fittings featured two holes just proximal and distal to the Sylgard tubes that were used to insert the cells for seeding. One of these holes was left open for gas exchange during the culturing process. The two three-way stopcocks on each fitting were used to switch the flow from the Sylgard tubes to a by-pass tube. During the startup period of the flow experiment, the fluid was directed through the by-pass until all air bubbles had left the system or dissipated. A reservoir (Fig. 1) provided fluid to the pump and allowed for gas exchange within the incubator. An Ismatech model MCP-Z (Ismatech S.A., Zurich, Switzerland) "steady flow" pump provided a relatively constant pressure. Following this pump was a ProMinent model gamma/5a (ProMinent Dosiertechnik GmbH, Heidelberg, Germany) diaphragm pump, which provided a 1 Hz pressure pulse. The two pumps were adjusted to provide a mean flow rate, onto which was superimposed a sinusoidal pulse flow rate. The mean and pulse flow rates are capable of being adjusted to simulate a range of physiologic conditions. For the results shown here, the flow rates through tubes b and d were adjusted to 0.54 1/min each, with an amplitude of 0.2/min. The flow rates were measured with two transit time ultrasonic flow probes (Tran-
Flow Control Valve
I•
Flow Rate Probe Diameter Probe
c
r3
Plastic Shell
d I DiaphragmI Pump I
D
FIGURE 1. Schematic of the flow system used to subject endothelial cells to shear stress and hoop stretch. The apparatus contains four compliant cylindrical tubes so that the following mechanical environments may be produced: (a) static conditions (low flow and a rigid jacket around the tube); (b) shear stress only (physiologic pulsatile flow and a rigid jacket); (c) hoop stretch only (low flow); and (d) shear stress and hoop stretch.
418
J.E. MOORE, JR.
et al.
O Compliant Tube
O
0
FIGURE 2. Illustration of the compliant tube onto which endothelial cells were cultured and the fittings in which it is mounted. (a) The pair of aluminum rods used to keep the tube straight and at a constant length. (b) Holes used to insert the cells for seeding. (c) Three-way valves used to switch the flow between the compliant tube and the bypass (d, complete tube not shown).
sonic Systems Inc., Ithaca, NY, U.S.A.). In order to monitor and adjust the hoop stretch, the diameters were measured near the centers of tubes c and d using the NIUS ultrasonic echo tracking device (Asulab, Neuchatel, Switzerland). This device is designed to measure dynamic changes in small artery diameters, and has an accuracy of approximately 3 microns (8). The thicknesses of the tubes, also measured with the NIUS device, were confirmed to not vary by more than 5% (10 microns) along the axis. The axial variation of the diameter waveform was confirmed to be no more than 5%. The mechanical design factors outlined above were taken into consideration in order to assure the viability of the ceils. Of equal importance are the biological techniques that were employed in order to maintain healthy monolayers of endothelial cells. The Sylgard tubes were hydrophyllized for 1 min in 70% sulfuric acid and rinsed extensively with deionized water. The tubes were mounted on the fittings as described above, gas sterilized (with ethylene oxide), and rinsed with sterile phosphate buffered saline (PBS). The threeway stopcocks were then switched to the by-pass position in order to shut off the passage of liquid from the Sylgard tubes. The insides of the tubes were coated with human fibronectin (Boehringer Mannheim, Mannheim, Ger-
many; 40 txg/ml in PBS) for 1 hr at room temperature and then rinsed with PBS. Calf pulmonary artery endothelial cells (CPAE, CCI 209) were purchased from the American Type Culture Collection (Rockville, MD, U.S.A.). These cells were then seeded at a density of 6-8 • 10 4 cells/cm 2 in four successive additions of 40 min. After seeding, the growth medium was added and cells were grown at 37~ under 5% CO2 with the fitting assembly vertically oriented to allow for uniform cell growth over the entire inside surface of the tube. Cells were routinely grown as static cultures in minimal essential medium with Earl's salts (EMEM, Seromed, Biochrom KG, Berlin), containing 20% fetal calf serum (FCS, Seromed) and supplemented with 10 mM HEPES (Seromed) and gentamycin (50 txg/ml). The cell monolayers reached confluence within two to three days, as ascertained under a light microscope. Finally, the seeding holes were capped with stainless steel screws. All parts of the flow system except the cell-cultured tubes and fittings were autoclaved and assembled under a laminar flow hood. Four straight tubes 30 diameters in length were inserted proximal to the cell-cultured tubes to ensure fully developed flow conditions. The flow medium was composed of EMEM, supplemented as the growth medium, but containing 10% FCS, 2% Rheomacrodex
Shearing and Stretching Endothelium (10% Dextran 40,000 in physiological saline, Pharmacia, Uppsala, Switzerland), and 1% human albumin (Swiss Red Cross, Central Laboratory, Bern, Switzerland). The kinematic viscosity of the resulting medium was 0.017 cm2/sec. Thus, the wall shear stress in tubes b and d was 8 --- 4 dynes/cm 2, as determined from the flow rate waveform using the method shown in He et al., (4). A hoop stretch of 6% (change in diameter divided by mean diameter) was generated in tube c, and 10% in tube d. These conditions were maintained for 24 hr. At the end of the flow experiment, the fittings were disconnected from the flow system and the cell monolayers were briefly inspected under an optical microscope. The medium was drained and the cells were rinsed once with PBS and fixed with 3% paraformaldehyde for 10 min at room temperature. The fixed tubes were disconnected from the fittings and sliced longitudinally. Once opened, the central part of each sheet was subdivided into two 10 mm x 15 mm rectangular pieces. In one piece of tubing, silver staining of the cell junctions was performed according to Zand et al. (10). Briefly, the fixed cells were washed twice in PBS, exposed to 0.25% AgNO3 for 1 min then rinsed twice in PBS. Silver cations were then complexed to bromide with 1% NHaBr/3% CoBr for 1 min. The cell ghosts were washed twice in PBS and exposed in 3% paraformaldehyde for 3-5 hr to light under a 60 Watt incandescent lamp in order to produce metallic silver that
419
deposits preferentially at intercellular junctions. Each Sylgard piece was then washed again in PBS, mounted onto a microscope slide and analyzed under a confocal scanning microscope (MRC500, Biorad, and Diaphot, Nikon) at 200 x or 400 x magnification. The cell shape index and alignment angle were determined according to Levesque and Nerem (6). The shape index indicates the degree of elongation of the cell, with a shape index of 1 for a circle, and a value of 0 indicating a straight line. The alignment angle was determined relative to the direction of flow (the axis of the tube). Following the PBS rinse of the second piece of tubing, the f-actin fibers within the fixed cells were labeled with flourescein-phalloidin, and their orientation was observed with confocal microscopy. RESULTS Following 24 hr of exposure to shear stress and/or hoop stretch, the morphology and f-actin structure of the endothelial cells from all four tubes were examined. The cells remained healthy and attached throughout the stressing process. It was found that the alignment and elongation of the cells was greater in tube d (shear stress and hoop stretch) than tube b (shear stress only). This alignment is evident from qualitative assessment of the silver staining images (Fig. 3). The shape indices and alignment angles of 40 cells in two randomly chosen areas near the center of the tube length were quantified in each of the four tubes.
FIGURE 3. Results of silver staining for intercellular junctions on cells exposed to the labeled mechanical stimuli (magnification = 2OOx). The flow direction was left to right, the pulsatile shear stress was 8 +- 4 dynes/cm 2 and the hoop stretch was 10% of the diameter for the tube subjected to both shear stress and hoop stretch. The cells exposed to combined shear stress and hoop stretch exhibited more alignment than those exposed to only shear stress.
420
J.E. MOORE, JR. et al.
In the static tube, the shape index averaged 0.874, with a standard error of the mean (SEM) of 0.019 (Fig. 4). For tube b (shear stress 8 --- 4 dynes/cm2), the shape index was 0.807 -+ 0.024 (p < 0.05 relative to control). For tube d (shear stress 8 • 4 dynes/cm z, hoop stretch = 10%), the shape index was 0.729 • 0.032 (p < 0.001 relative to control; p = 0.05 relative to tube b). The alignment angle was 61.5 • 4.1, 36.6 +- 4.8, and 16.2 • 2.9 degrees for the control, and tubes b and d, respectively (t7 < 0.001 for all comparisons). The standard deviation of the alignment angle for the control cells was 26 ~ indicating a large dispersion of values, as would be expected for randomly oriented round cells. For tube c (hoop stretch = 6%), the shape indices and the orientation angle were not statisti1.0
X Q "o I=
0.8
q) D. r J~ U)
0.6
Static
Shea?stress
Shear stress and Hoop stretch
(a)
80 Q
9
60
Q
~
40
e= < C 0
20
o
o
Static
Shear'stress
Shea~rstress and Hoop stretch
(b) FIGURE 4. Quantitative results of the shape indices and alignment angles of 40 of the cells shown in each part of Fig. 3. Error bars represent the standard error of the mean (SEM). According to the shape index results, the cells exposed to shear stress elongated more than those in static culture {p < 0,05), while those exposed to both shear stress and hoop stretch elongated slightly more than those exposed to shear stress only (p = 0.05). The alignment angles were closer to the axis of the tube for the cells exposed to shear stress than control (p < 0.001), and the cells exposed to shear stress and hoop stretch aligned more than those exposed to only shear stress (p < 0.001).
cally different from the control (mean shape index = 0.89 - 0.016, mean alignment angle = 54.7 -+ 4.2~ Results from the labeling of f-actin indicated that the fibers in tube d were more oriented with the axis of the tube than in tubes b and c (Fig. 5). Dense peripheral bands of f-actin were seen in the static culture (tube a). The cells in tube c, although not seen to elongate significantly, nevertheless had begun to reorganize their f-actin structure to align with the axis of the tube. DISCUSSION An experimental apparatus was designed with the intention of subjecting vascular endothelial cells to simultaneous physiologic levels of both fluid shear stress and cyclic circumferential " h o o p " stretch. Numerous mechanical and biological considerations were taken into account in the design of the apparatus. The device was capable of producing physiological levels of both shear stress and hoop stretch. Measurements of the diameter waveforms along the length of the tubes demonstrated that the hoop stretch was uniform throughout each tube. The cells remained viable for the entire experiment, and exhibited different morphological responses to each of the four mechanical environments. The organization of the f-actin network was also observed qualitatively to be affected by different stress environments. The axial alignment of cells subjected to shear stress and hoop stretch was greater than that seen for cells subjected to shear stress only. The alignment of the cells in response to shear stress is in general agreement with the alignment noted in previous studies of cells cultured on flat rigid plates (6). The enhanced alignment of the cells in tube d with hoop stretch is also in agreement with previous studies of the effects of cyclic stretching of flat compliant membranes, which observed alignment of the cells perpendicular to the stretch direction (5). It is important to note that those previous studies were separate; no previous method for subjecting vascular cells to combined physiologic levels of shear stress and hoop stretch existed, to the authors' knowledge. Another advantage of this flow system over previously reported "fiat-plate" arrangements is that the cells that are meant to be exposed to only hoop stretch are subjected to a very low, spatially constant shear stress ( < 0 . 5 dyne/cm2), rather than spatially varying oscillating shear stresses. In the experimental apparatus of Ives et al. (5), a flat rectangular compliant substrate was clamped at one end, and stretched from the other. This means that the cells are subjected to spatially varying shear stresses, in a similar fashion to Stokes' second problem (7). The shear stress near the clamped end is zero, while the shear stress at the other end depends on the amount of stretch. Although this shear stress was reported to be less than 1 dyne/cm 2 at 10% elongation, it was apparently sufficient to alter the
Shearing and Stretching Endothelium
421
FIGURE 5. Results of f-actin labeling with flourescein-phalloidin in cells exposed to the labeled mechanical stimuli (magnification = 600x). The flow direction was left to right, the puIsatile shear stress was 8 -+ 4 dynes/cm 2 and the hoop stretch was 10% of the diameter for the tube subjected to both shear stress and hoop stretch. The f-actin in the cells exposed to both shear stress and hoop stretch showed more enhanced organization in the direction of the axis of the tube than cells exposed to only shear stress.
rates of production of endothelin and prostacyclin (3). This indicates that the spatially varying oscillating shear stress that results from the stretching of the substrate does in fact alter the cells' behavior in addition to the stretch itself. Other apparati designed to subject cells to cyclic stretch include flexible-bottom culture plate systems (9). The advantage of the flow system presented here over such devices is that the stretch is spatially constant and uniaxial (all circumferential), rather than spatially varying and biaxial (radial and circumferential). In summary, it can be said that the apparatus met all of the mechanical and biological criteria. A stretch of only 6% was attainable in tube c for the particular results shown because of the need to limit the flow rate through this tube. This lowered the pressure in tube c considerably, relative to tube d. In subsequent mechanical trials, a thinner walled tube was used for tube c, and larger hoop stretches were attained. Since the amount of stretch is the target parameter, not the actual compliance of the tube, varying the wall thickness is a valid method of obtaining a wider range of mechanical stimuli. The lower level of hoop stretch was most likely responsible for the cells in tube c not aligning significantly more than the control cells. This did not detract from the observation of enhanced alignment of cells subjected to both shear stress and hoop strain relative to those subjected to only shear stress. There was
apparently no contamination of the flow system during the experiment, and the cells were able to adhere to the tubes and survive attached under shear stress and hoop stretch. Future experiments will focus on a number of cell function parameters under different shear stress and hoop stretch conditions. Eventually, it should be possible to determine which type of physiological mechanical stimulus is more important in determining the behavior of vascular endothelial cells. This work represents a significant step in examining the effects of mechanical phenomena on endothelial cell behavior in that both types of stress to which the cells are actually subjected are included. The enhanced alignment of the cells due to the addition of hoop stretch to shear stress had not previously been observed. It is important to note that these results are preliminary in nature and only represent the results of a single experiment. More results are necessary in order to draw substantial conclusions. The results from future work with this apparatus will lead to a better understanding of the relationship between mechanical factors and cellular physiology, as well as the relevance to atherogenesis. REFERENCES
1. Banes, A.J. Mechanical strain and the mammalian cell. In: Frangos, J.A., ed. Physical forces and the mammalian cell. New York: Academic Press; 1993: pp. 81-123.
422
J.E. MOORE, JR. et al.
2. Berthiaume, F.; Frangos, J.A. Effects of flow on anchorage-dependent mammalian cells-secreted products. In: Frangos, J.A., ed. Physical forces and the mammalian cell. New York: Academic Press; 1993: pp. 139-192. 3. Carosi, J.A.; Eskin, S.G.; Mclntire, L.V. Cyclical strain effects on production of vasoactive materials in cultured endothelial cells. J. Cell. Physiol. 151:29-36; 1992. 4. He, X.; Ku, D.N.; Moore, J.E.; Jr. Simple calculation of the velocity profiles for pulsatile flow in a blood vessel using Mathematica. Ann. Biomed. Eng. 21:45-49; 1993. 5. Ives, C.L.; Eskin, S.G.; Mclntire, L.V. Mechanical effects on endothelial cell morphology: in vitro assessment. In Vitro Cell. Dev. Biol. 22(9):50(0507; 1986. 6. Levesque, M.J.; Nerem, R.M. The elongation and orientation of cultured endothelial cells in response to shear stress. ASME J. Biomech. Eng. 107:341-347; 1985.
7. Panton, R.L. Incompressible flow. New York: John Wiley & Sons; 1984. 8. Tardy, Y.; Meister, J.J.; Perret, F.; Brunner, H.R.; Arditi, M. Noninvasive estimate of the mechanical properties of peripheral arteries from ultrasonic and photoplethysmographic measurements. Clin. Phys. Physiol. Meas. 12(1): 39-54; 1991. 9. Williams, J.L.; Chen, J.H.; Belloli, D.M. Strain fields on cell stressing devices employing clamped circular elastic diaphragms as substrates. ASME J. Biomech. Eng. 114: 377-384; 1992. 10. Zand, T.; Underwood, J.M.; Nunnari, J.J.; Majno, G.; Joris, I. Endothelium and "silver lines." An electron microscopic study. Virchows Arch. Pathol. Anat. 395:133144; 1982.
